INTRODUCTION OF RESEARCH TOPICS AND CURRENT RESEARCHES IN INSURANCE 박소정 ( 서울대학교 ), 금융경제연구원세미나

Transcription

1 INTRODUCTION OF RESEARCH TOPICS AND CURRENT RESEARCHES IN INSURANCE 박소정 ( 서울대학교 ), 금융경제연구원세미나

2 My Current Working/Published Papers Research interest: Anything related to Insurance and Risk Management! Research papers 1. Insurance Company Operations Real Effects of Rating Standards for Catastrophic Risks Reinsurance and Systemic Risk: The Impact of Reinsurer Downgrading on Property-Casualty Insurers The Opaqueness of Structured Bonds: Evidence from the US Insurance Industry The Opacity of Insurance Companies 2. Insurance Demand The Impact of Culture on the Demand for Non-Life Insurance (published) Culture Matters: Long-Term Orientation and Life Insurance Demand (forthcoming) Do Japanese Policyholders Care About Insurers Credit Quality? (forthcoming) 3. Actuarial Science Is the Design of Bonus-Malus Systems Influenced by Insurance Maturity or National Culture? (published) The Use of Annual Mileage as a Rating Variable

3 Insurance 101 Insurance Risk transferring Risk pooling (Law of Large Numbers!) Insured σ σ N Insurance Company σ σ σ σ

4 Insured Catastrophe Losses

5 Catastrophic Losses Reinsurance Alternative Risk Transfer (ART) the convergence of insurance and financial market CAT Bonds mid-1990s after Andrew/Northridge, 14% after 911, 20% after Katrina Sidecars (disposable reinsurers) After 911 & Katrina Easy start up & built-in exit Regulation Solvency Regulation Guarantee Fund policyholder protection State/Federal Insurance California: State-owned insurance company (CEA) Terrorism: Federal reinsurance program (TRIA) Florida/Louisiana: Residual market (Citizens Insurance Company) Moral Hazard issues... Insurance company Insured

6 Reinsurance insured insured Insurance Company Reinsurance Company insured Insurance Company Insurance Company Reinsurance Company Retrocession

7 Catastrophe Bonds -Bond with multiple risk levels $100 Million Cat Bond Tranche C $25 million, Moody s Rated A LIBOR bps Tranche B $50 million, Moody s Rated Baa LIBOR +400 bps Tranche A $25 million, Moody s Rated B LIBOR bps 0.25% Chance of Loss 1% Chance of Loss 3% Chance of Loss

8 Catastrophe Bond Issuance 2010: 23 issues with volume $4.8b (Aon Benfield Securities report)

9 Typical Sidecar Transaction Structure 9

10 Example Sidecar Financial Responsibility Spectrum 10

11 11 Real Effects of Rating Standards for Catastrophic Risks -preliminary draft- Anastasia V. Kartasheva (Bank of International Settlement) Sojung C. Park (Seoul National University)

12 12 Number of insurers in hurricane risk exposed states by lines of business and A.M. Best Rating Year Commercial Strong Commercial Weak Total Commercial Personal Strong Personal Weak Total Personal Weak rating: A- and lower or not rated Commercial insurer: more than 50% of business written in commercial lines of business

13 Rating Change 13 Higher Rating 11.1 Mean Rating of CAT PC insurers (Tracking same insurers) Commercial Personal Lower Rating

14 Rating Trend Personal Lines 14 Total Direct Premium Written of personal line PC insurers in CAT states by A.M. Best Rating (Whole sample) 천 Personal High Personal Low

15 Background: Rating Standard Change 15 Standard & Poor's capital charge for catastrophe risk is based on the company-specific net expected losses at the 1-in-250-year level. and the 1-in-250-year charge is for natural catastrophes within the property line of business only. The required capital needed at a 'BBB' level of confidence is the modeled 1-in-250- year scenario level. Previously, a 1-in-100-year capital charge was applied to some reinsurance companies, but according to this updated criteria, a 1-in- 250-year level charge will be applied to all reinsurers globally. The basis for the 1-in-250-year versus a 1-in-100-year charge is a recognition by Standard & Poor's that catastrophe losses in general are increasing from the perspectives of both frequency and severity. (Source: Reinsurance Criteria: Larger Losses And Better Modeling Prompt Changes To Property Catastrophe Criteria, Standard & Poors Rating Direct, 2005)

16 Research question 16 Natural Experiment: In the major rating agencies has adjusted the criteria for insurers with e xposure to catastrophic events In 2005, CRAs introduces new standards to reinsurance In 2006, the changes were extended to major insurers Has insurance companies increased the amount of c apital as a result of more stringent rating standard change? Not necessarily! It is heterogeneous.

17 17 Ratings and Capital: Why ratings matter? Traditional theories of capital structure do not incorporate credit rating as a decision factor. Rating in general Regulations on bond investment are based directly on credit ratings Rating provides signals of financial quality of firms to investors Rating in insurance industry Demand for insurance is sensitive to insurer s financial quality Buyers are willing to pay higher price for insurance sold by higher quality company Insurance buyers rely on ratings to assess insurers financial quality Discrete costs (benefits) associated with different ratings level! Similar firms are pooled in the same rating bin.

18 18 Ratings and Capital: Why ratings matter? Kisgen (JF, 2006) Test the impact of credit ratings on capital structure decisions. Firms near a change in rating issue less net debt relative to net equity over a subsequent period compared to other firms. Our contribution: Test the impact of credit rating on capital structure decision using natural experiment Test the impact of rating standard change and find heterogeneous behavior

19 Heterogeneous reaction to new standards 19 Ratings pool different quality companies in rating gr ades Old standard B A New standard B A Credit quality These companies are do wngraded to rating B un der new standard These companies maint ain rating A under new standard

20 Heterogeneous reaction to new standards 20 Two possible reactions 1. Increase credit quality to maintain rating A 2. Accept the downgrade to B (and possible decrease credit qualit y) The choice between #1 and #2 depends on Sensitivity of demand to credit quality Cost of capital Old standard B A New standard B A Credit quality What these companies are going to do?

21 Heterogeneous Propensity 21 Their propensity to defend a rating (Epermanis and Harrington (2006), Nini and Chyene (2010) ) Line of business: commercial insurer vs. personal line insu rer Rating: A-

22 Testing Hypotheses 22 Rating standard change and capital change Asymmetric reaction to the standard change 1) Firms closer to lower boundary of each rating bin will ch ange their surplus level more 2) change : either up or down Commercial insurers: more likely to move up Commercial insurers with A- rating: more likely to move up Personal line insurers: may reduce surplus level if it is too costly

23 Data 23 Sample: All PC insurers with A.M. Best s rating: Firm - year number of observations: 10,910 CAT exposed sample: PC insurers with positive premium writte n In homeowners, farmowners, auto physical damage, commercial mult iperil, inland marine in AL, FL, MS, SC, or TX Firm year number of observations: 3,055 Data source A.M. Best rating guide NAIC (National Association of Insurance Commissioners)

24 Methodology 24

25 Empirical analysis strategy Finding Rating Model: ordered probit model a. Estimate a and b Rating (numerical conversion: 1 to 13*) = a + bx b. Get estimates of Z the continuous latent variable Z = a + bx * A++: 13, A+: 12, A: 11,.

26 Empirical analysis strategy 26 X: Explanatory variables: BCAR (Best s Capital Adequacy Ratio) Combined Ratio Premium Surplus ratio Public company dummy Proportion of CAT line of business DPW to total DPW And etc. Goal: Estimating the latent variable Z Only keep statistically significant X variables. Coefficients robust to excluding insignificant variables.

27 Empirical analysis strategy Estimate the distance: Step 1. Estimate upper boundary Z for each rating bin, year Step 2. DIS i = Upper Boundary Z - Z i Larger Distance: closer to lower boundary

28 Empirical analysis strategy Construct CATDIS variable CATDIS05= DIS * Proportion of Premium Written in Catastrophic Risk Exposed Line of Business in 2005 CATDIS06= DIS * Proportion of Premium Written in Catastrophic Risk Exposed Line of Business in 2006

29 Empirical analysis strategy Evaluate the change in capital structure depending on the position of the company in the rating bin, con trolling for demand sensitivity Surplus Asset ratio change (t, t+1) = a + b1* CATDIS05 +b2*catdis06+c X

30 Results 30

31 Step 1: Ordered Probit Model 31 Rating = a + bx A++=13, A+=12, A=11,.

32 32 Variables Whole Sample Investment Yield.0436 [.0060] *** Combined Ratio [.0002] *** Premium/Surplus [.0002] *** Net Premium Growth.0005 [.0000] *** Reinsurance Recoverable/Surplus [.0001] *** Reserve/Surplus [.0002] *** Junk Bonds/Asset [.0018] *** BCAR.0010 [.0000] *** Log[Asset].3721 [.0070] *** Public Dummy.5836 [.0236] *** Single Dummy [.0314] *** Firm Age.0010 [.0003] *** PCAT [.0707] *** Likelihood Ratio[LR] 5,551.3 Estrella pseudo-r McFadden's LRI Number of Obs 10,910

33 Ordered Probit Model 33 Rating Standard Change on CAT risk and Rating? PCAT: Proportion of CAT line of business to total DPW Regressions two subsample: Before change: After change:

34 34 Variables Investment Yield [.0070] *** [.0123] *** Combined Ratio [.0002] ** [.0003] *** Premium/Surplus [.0002] *** [.0003] *** Net Premium Growth [.0001] *** [.0002] Reinsurance Recoverable/Surplus [.0001] *** [.0002] *** Reserve/Surplus [.0002] *** [.0003] *** Junk Bonds/Asset [.0020] *** [.0045] *** BCAR [.000] *** [.0001] *** Log[Asset] [.0089] *** [.0123] *** Public Dummy [.0290] *** [.0401] *** Single Dummy [.0392] *** [.0529] *** Firm Age [.0003] *** [.0004] * PCAT [.0873] *** [.1180] *** Likelihood Ratio[LR] 3, ,900.7 Estrella pseudo-r McFadden's LRI Number of Obs 7,118 3,792

35 Step 3: Regression Model l 35 Test whether those insurers close to the bottom of each rating bin and insure rs more sensitive to the rating change increase their surplus more than others in Surplus Asset ratio change (t, t+1) = a + b1 CATDIS05 +b2 CATDIS06 + c X Asymmetric reaction If firms closer to lower boundary reduce (increase) capital more than other firms in 2005 and 2006, b1 and b2 expected to be negative (positive)

36 36 By A.M. Best Rating in year t Commercial Insur Variables All PC Insurers ers Personal Insurers Intercept *** *** *** [.045] [.047] [.104] DISTANCE.0199 ***.0182 ***.0279 *** [.002] [.002] [.004] CATDIS *** *** [.007] [.022] [.008] CATDIS * [.010] [.029] [.011] CAT *** [.015] [.047] [.018] CAT [.022] [.067] [.025] Log[Asset].0514 ***.0426 ***.0758 *** [.003] [.004] [.009] Δ Combined Ratio *** *** *** [.000] [.000] [.000] Δ RBC [.000] [.000] [.000] Pseudo R-Square F-value Number of Obs. 3,055 2,

37 37 N R-Square CATDIS05 CATDIS06 Commercial A and above 1, A *** B++ and below Personal A and above ** A B++ and below ***

38 Conclusion 38 Rating matters for corporate finance decisions The effect is heterogeneous Sensitivity of insurance demand with respect to credit quality Commercial insurers vs. Personal insurers Differences across the rating scale A- rated commercial insurer Reaction of personal insurers Stringent rating standard change does not necessarily im prove the financial quality of all insurers.

39 The Impact of Culture on the Demand for Non-Life Insurance Sojung Park (Seoul National University) Jean Lemaire (University of Pennsylvania)

40 Culture and Insurance Demand Log(Penetration) Log(GDP)

41 Culture and Insurance Demand Log(Penentration) Log(GDP)

42 Hofstede Cultural Measures Power distance (PDI) Individualism (IDV) Masculinity (MAS) Uncertainty avoidance (UAI) Long-term orientation (LTO) (Not included)

43 Results (1) (2) (3) (4) (5) (6) Developing Countries Developed Countries log GDP 0.496*** 0.445*** 0.379*** *** *** *** (0.082) (0.079) (0.095) (0.054) (0.054) (0.052) URBAN * (0.232) (0.220) (0.244) (0.139) (0.145) (0.115) EDUC *** *** *** 2.567*** 2.121*** (0.493) (0.514) (0.525) (0.403) (0.420) (0.392) PRISK 0.092*** 0.076*** 0.060** 0.167*** 0.116*** 0.060*** (0.023) (0.027) (0.029) (0.018) (0.019) (0.014) log HERF *** *** *** *** *** *** (0.033) (0.027) (0.033) (0.018) (0.017) (0.018) COMMON 0.174*** (0.064) (0.100) (0.050) (0.029) ISLAM *** *** *** *** (0.108) (0.101) (0.101) (0.117) IDV *** (0.233) (0.107) UAI *** (0.196) (0.074) PDI *** *** (2.105) (0.858) MAS ** (0.253) (0.054) Constant *** *** *** 4.829*** 2.946*** 1.893*** (0.664) (0.650) (0.795) (0.517) (0.513) (0.512) Observations Adjusted R-square F Value

44 Conclusion Culture matters in developed economy